A fast, effective data attribution method for neural networks in PyTorch

{KFAC,EKFAC,Diagonal,Implicit} Fisher Matrices and finite width NTKs in PyTorch

Multi-framework implementation of Deep Kernel Shaping and Tailored Activation Transformations, which are methods that modify neural network models (and their initializations) to make them easier to train.

The first open-source repository dedicated to PINN research via Vibe Coding. Complete JAX-GPU implementations of PINN algorithms with Chinese tutorials. 首个采用 Vibe Coding 进行 PINN 研究的开源仓库。

We propose a lossless compression algorithm based on the NTK matrix for DNN. The compressed network yields asymptotically the same NTK as the original (dense and unquantized) network, with its weights and activations taking values only in {0, 1, -1} up to scaling.

Official repository of our work "Finding Lottery Tickets in Vision Models via Data-driven Spectral Foresight Pruning" accepted at CVPR 2024

Neural Tangent Kernel (NTK) module for the scikit-learn library

Code accompanying the paper "On the adaptation of recurrent neural networks for system identification"

Official Code: Trust Your Robots! Predictive Uncertainty Estimation of Neural Networks with Sparse Gaussian Processes

TF2 Implementation of Physics Informed Neural Networks and Neural Tangent Kernel

TCT: Convexifying Federated Learning using Bootstrapped Neural Tangent Kernels

Code for "Learnware Reduced Kernel Mean Embedding Specification Based on Neural Tangent Kernel"

Composable, differentiable, and simple finite-width and infinite-width NTKs for Gaussian Processes, with cross-functionality with GPyTorch.

Interactive spectral geometry visualization on MNIST - companion code for The Spectral Structure of Natural Data

Implementation of Approximate Smooth Kernel Value Iteration

🧮 Explore vector operations interactively with Python, covering dot products, cross products, and visualizations for enhanced understanding.